Research on Self Inspection Algorithm of Field Operation Based on Improved YOLOV8n

Aiming at the problems of small target size, complex occlusion and difficult deployment of edge equipment in the task of autonomous inspection on the operation site, a lightweight detection algorithm based on improved yolov8n is proposed. Firstly, the c2fghost DCCA feature extraction module is constructed, which combines dynamic convolution and channel attention mechanism to solve the problem of feature redundancy under complex background interference; On this basis, a bidirectional feature pyramid structure is designed to enhance the small target detection ability through multi-scale feature interaction; Then the asymmetric decoupling detection head is used to optimize the local feature expression, and the improved IOU loss function is combined to improve the accuracy of boundary box regression. Experimental results show that compared with the mainstream models such as yolov8n and yolov10n, the proposed algorithm improves map by 6.2% to 94.1% and reduces the number of parameters by about 22.5%. This research not only provides a new method for small target real-time detection in the field operation scene, but also provides technical support for the lightweight model of edge computing device deployment, which has engineering practical value.